liaopeiyuan / kaggle-solutions

Solutions to my previous Kaggle competitions

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

kaggle-solutions

(Updated Sep 2023) This is a repository containing some of my solutions to Kaggle competitions from 5 years ago. It is no longer being maintained. Feel free to use them as you wish.

Dependencies

  • tensorflow-gpu
  • torch
  • torchvision
  • tensorboard
  • Pillow
  • opencv-python
  • pandas
  • numpy

Past Collaborators

TGS: Renan @scherzocrk, Ding Han @DhDiego

Doodle: Venn @wangvenn, SeuTao @SeuTao, earhian @earhian, Tom57 @ThomasTilli, CHAN, wilan @wilanw, Andi Leslie Yan @RedLiN5

Current projects:

  1. TGS Salt Identification Challenge
model #fold #models public LB private LB placement
ResNet34-512 1 1 0.859 0.862 N/A
ResNet34-256 1 1 0.868 0.870 N/A
ResNet34-128 1 1 0.837 0.855 N/A
ResNet34-128 1 6 0.854 0.873 N/A
ResNet34-128 10 60 0.869 0.884 N/A
SE-ResNeXt-50-BN-128 3 3 0.861 0.880 N/A
ResNet34-128+256-post 1 2 0.875 0.888 22
SE-ResNeXt-50-OC-128 5 15 0.869 0.889 20
SE-ResNeXt-50-OC-256-post 10 95 0.881 0.892 9
ResNet34-256 10 60 0.875 0.894 8
2models-2scale-stable-post 10 100 0.885 0.894 5
  1. Quick, Draw! Doodle Recognition Challenge

About

Solutions to my previous Kaggle competitions

License:Apache License 2.0


Languages

Language:Python 67.4%Language:Jupyter Notebook 32.6%Language:Lua 0.0%Language:Shell 0.0%